As the measurement terminal near the user side in the smart grid, the smart meter is the key parts of the interaction with the user in the smart grid advanced metering infrastructure, therefore, its reliability needs to be focused. Aiming at the shortcomings of existing smart meter core components selection method based on the subjective experience, this paper proposes an automatic selection method for the key components of the smart meter, which overcomes empirical selection method that rely on subjective experience heavily. Firstly, in this paper, the analytic hierarchy process is used to select components in smart meters. Secondly, for the shortcoming of objective weight of analytic hierarchy process that relies too much expert experience, this paper uses the entropy weight method to further adaptively adjustment the weight of the indicator. Finally, through the selection experiment of key components in the smart meter, the feasibility and effectiveness of the proposed selection method is verified, which can guarantee the reliable measurement for the smart meter.